A cross-domain fruit classification method based on lightweight attention networks and unsupervised domain adaptation
Abstract Image-based fruit classification offers many useful applications in industrial production and daily life, such as self-checkout in the supermarket, automatic fruit sorting and dietary guidance. However, fruit classification task will have different data distributions due to different applic...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2022-12-01
|
| Series: | Complex & Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s40747-022-00955-8 |
| _version_ | 1797769244352446464 |
|---|---|
| author | Jin Wang Cheng Zhang Ting Yan Jingru Yang Xiaohui Lu Guodong Lu Bincheng Huang |
| author_facet | Jin Wang Cheng Zhang Ting Yan Jingru Yang Xiaohui Lu Guodong Lu Bincheng Huang |
| author_sort | Jin Wang |
| collection | DOAJ |
| description | Abstract Image-based fruit classification offers many useful applications in industrial production and daily life, such as self-checkout in the supermarket, automatic fruit sorting and dietary guidance. However, fruit classification task will have different data distributions due to different application scenarios. One feasible solution to solve this problem is to use domain adaptation that adapts knowledge from the original training data (source domain) to the new testing data (target domain). In this paper, we propose a novel deep learning-based unsupervised domain adaptation method for cross-domain fruit classification. A hybrid attention module is proposed and added to MobileNet V3 to construct the HAM-MobileNet that can suppress the impact of complex backgrounds and extract more discriminative features. A hybrid loss function combining subdomain alignment and implicit distribution metrics is used to reduce domain discrepancy during model training and improve model classification performance. Two fruit classification datasets covering several domains are established to simulate common industrial and daily life application scenarios. We validate the proposed method on our constructed grape classification dataset and general fruit classification dataset. The experimental results show that the proposed method achieves an average accuracy of 95.0% and 93.2% on the two datasets, respectively. The classification model after domain adaptation can well overcome the domain discrepancy brought by different fruit classification scenarios. Meanwhile, the proposed datasets and method can serve as a benchmark for future cross-domain fruit classification research. |
| first_indexed | 2024-03-12T21:06:09Z |
| format | Article |
| id | doaj.art-8fad712effd648b686f533c544c6a08c |
| institution | Directory Open Access Journal |
| issn | 2199-4536 2198-6053 |
| language | English |
| last_indexed | 2024-03-12T21:06:09Z |
| publishDate | 2022-12-01 |
| publisher | Springer |
| record_format | Article |
| series | Complex & Intelligent Systems |
| spelling | doaj.art-8fad712effd648b686f533c544c6a08c2023-07-30T11:28:18ZengSpringerComplex & Intelligent Systems2199-45362198-60532022-12-01944227424710.1007/s40747-022-00955-8A cross-domain fruit classification method based on lightweight attention networks and unsupervised domain adaptationJin Wang0Cheng Zhang1Ting Yan2Jingru Yang3Xiaohui Lu4Guodong Lu5Bincheng Huang6State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang UniversityState Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang UniversityState Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang UniversityState Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang UniversityState Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang UniversityState Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang UniversityKey Laboratory of Cognition and Intelligence Technology, China Electronics Technology Group CorporationAbstract Image-based fruit classification offers many useful applications in industrial production and daily life, such as self-checkout in the supermarket, automatic fruit sorting and dietary guidance. However, fruit classification task will have different data distributions due to different application scenarios. One feasible solution to solve this problem is to use domain adaptation that adapts knowledge from the original training data (source domain) to the new testing data (target domain). In this paper, we propose a novel deep learning-based unsupervised domain adaptation method for cross-domain fruit classification. A hybrid attention module is proposed and added to MobileNet V3 to construct the HAM-MobileNet that can suppress the impact of complex backgrounds and extract more discriminative features. A hybrid loss function combining subdomain alignment and implicit distribution metrics is used to reduce domain discrepancy during model training and improve model classification performance. Two fruit classification datasets covering several domains are established to simulate common industrial and daily life application scenarios. We validate the proposed method on our constructed grape classification dataset and general fruit classification dataset. The experimental results show that the proposed method achieves an average accuracy of 95.0% and 93.2% on the two datasets, respectively. The classification model after domain adaptation can well overcome the domain discrepancy brought by different fruit classification scenarios. Meanwhile, the proposed datasets and method can serve as a benchmark for future cross-domain fruit classification research.https://doi.org/10.1007/s40747-022-00955-8Transfer learningDeep learningAttention mechanismFruit classification |
| spellingShingle | Jin Wang Cheng Zhang Ting Yan Jingru Yang Xiaohui Lu Guodong Lu Bincheng Huang A cross-domain fruit classification method based on lightweight attention networks and unsupervised domain adaptation Complex & Intelligent Systems Transfer learning Deep learning Attention mechanism Fruit classification |
| title | A cross-domain fruit classification method based on lightweight attention networks and unsupervised domain adaptation |
| title_full | A cross-domain fruit classification method based on lightweight attention networks and unsupervised domain adaptation |
| title_fullStr | A cross-domain fruit classification method based on lightweight attention networks and unsupervised domain adaptation |
| title_full_unstemmed | A cross-domain fruit classification method based on lightweight attention networks and unsupervised domain adaptation |
| title_short | A cross-domain fruit classification method based on lightweight attention networks and unsupervised domain adaptation |
| title_sort | cross domain fruit classification method based on lightweight attention networks and unsupervised domain adaptation |
| topic | Transfer learning Deep learning Attention mechanism Fruit classification |
| url | https://doi.org/10.1007/s40747-022-00955-8 |
| work_keys_str_mv | AT jinwang acrossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT chengzhang acrossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT tingyan acrossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT jingruyang acrossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT xiaohuilu acrossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT guodonglu acrossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT binchenghuang acrossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT jinwang crossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT chengzhang crossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT tingyan crossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT jingruyang crossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT xiaohuilu crossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT guodonglu crossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation AT binchenghuang crossdomainfruitclassificationmethodbasedonlightweightattentionnetworksandunsuperviseddomainadaptation |